LST-SLAM: A Stereo Thermal SLAM System for Kilometer-Scale Dynamic Environments

2026-02-24

AI Summary

This work proposes LST-SLAM, a stereo thermal-inertial SLAM system designed for kilometer-scale dynamic outdoor environments, addressing challenges such as unreliable thermal features, unstable tracking, and inconsistent mapping. The approach integrates self-supervised thermal feature learning, a two-level stereo motion tracking framework, and geometric pose optimization. To mitigate dynamic disturbances, it incorporates semantic-geometric hybrid constraints and employs an online incremental bag-of-words model for efficient loop closure detection. Evaluated on large-scale dynamic thermal datasets, LST-SLAM demonstrates significantly improved robustness and localization accuracy compared to state-of-the-art methods AirSLAM and DROID-SLAM, confirming its effectiveness and technical advancement.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Motion & Tracking
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Thermal cameras offer strong potential for robot perception under challenging illumination and weather conditions. However, thermal Simultaneous Localization and Mapping (SLAM) remains difficult due to unreliable feature extraction, unstable motion tracking, and inconsistent global pose and map construction, particularly in dynamic large-scale outdoor environments. To address these challenges, we propose LST-SLAM, a novel large-scale stereo thermal SLAM system that achieves robust performance in complex, dynamic scenes. Our approach combines self-supervised thermal feature learning, stereo dual-level motion tracking, and geometric pose optimization. We also introduce a semantic-geometric hybrid constraint that suppresses potentially dynamic features lacking strong inter-frame geometric consistency. Furthermore, we develop an online incremental bag-of-words model for loop closure detection, coupled with global pose optimization to mitigate accumulated drift. Extensive experiments on kilometer-scale dynamic thermal datasets show that LST-SLAM significantly outperforms recent representative SLAM systems, including AirSLAM and DROID-SLAM, in both robustness and accuracy.

Problem

• thermal SLAM
• dynamic environments
• feature extraction
• motion tracking
• large-scale

Innovation

• thermal SLAM
• self-supervised feature learning
• stereo motion tracking
• semantic-geometric constraint
• incremental bag-of-words